Optimization of seaweeds biomass via machine learning models for bioethanol production: Biological pretreatment and total reducing sugar

Seaweeds (SWs) represent a promising feedstock for biofuel production due to rich in polysaccharides, which are suitable substrates for fermentation into bioethanol. However, the complex polysaccharides in SWs require effective pretreatment to break down into simple sugars. Despite this, few studies have explored bacterial consortia as a pretreatment to enhance total reducing sugar (TRS) yield. Thus, the current study focused on the biological pretreatment of SWs, along with the prediction of bioethanol via artificial neural networks (ANN) and extreme gradient boosting (XGBoost). The bacterial pretreatment using a consortium of Pseudomonas and Bacillus species significantly increased the TRS level to 99.489, 117.373, and 133.241 mg/g biomass of Grateloupia subpectinata GEEL-22, Sargassum muticum GEEL-25, and Sargassum stenophyllum GEEL-38, respectively. During separate hydrolysis and fermentation (SHF) using both free and immobilized yeast cells, S. stenophyllum GEEL-38 exhibited the highest bioethanol production and sugar utilization, reaching 3.15 g/L with 83.2% and 3.41 g/L with 84.8%, respectively. Principal component analysis (PCA) revealed strong correlations among pH levels, sugar utilization, and bioethanol yield, highlighting the importance of maintaining optimal pH and maximizing sugar consumption in fermentation. Additionally, ANN and XGBoost models demonstrated coefficients of determination (R 2 ) of 0.87 and 0.78, respectively, with the ANN model showing superior performance as indicated by a mean squared error (MSE) of 0.1093. The integration of bacterial pretreatment with machine learning models offers a promising strategy for enhancing bioethanol production from SWs, which can be further implemented for industrial bioethanol production.